The problem with doomsday dropout data.

A new consensus says we have a high school dropout crisis with about one-third of all students and half of all minority students failing to graduate. Jay Greene of the Manhattan Institute and Christopher Swanson, formerly of the Urban Institute, have published widely cited research supporting this view.

Their estimates seem straightforward. Although their methods differ slightly, their approaches are similar: Obtain the number of diplomas issued in a year from the U.S. Department of Education. Then divide this number by 9th-grade enrollment three years earlier to get a graduation rate.

This method is flawed. One problem is the 9th-grade "bulge." Many students spend more than one year as freshmen, usually because they fail a 9th-grade course and must retake the class during the following year. In public schools nationwide, there are about 13 percent more 9th graders than there were 8th graders the previous year. For blacks and Hispanics, the difference is closer to 25 percent. Thus dividing diplomas awarded (the numerator) by 9th graders three years earlier (the denominator) yields an artificially low graduation rate because the denominator has been exaggerated.

Imprecise Measurements

Proponents of the crisis view make only a halfhearted attempt to correct this problem. For example, in recent publications, Greene averages 8th, 9th and 10th grade enrollment for his denominator. But if the 9th-grade number is bloated, averaging it with 8th and 10th grade enrollment will not eliminate the problem. The remaining retention bulge in Greene's data incorrectly lowers the black graduation rate by 5 percent and the Hispanic rate by 7 percent.

Greene's estimates also are confounded by immigration. If a 16-year-old Mexican immigrant enrolls in high school and graduates, this would make the graduation rate seem artificially high because that youth was not counted in earlier 9th-grade enrollment. So Greene simply inflates his denominator (9th grade enrollment) by the total of 15-, 16- and 17-year-old immigrants in the next three years. He implicitly assumes every teenage immigrant enters high school and that those not graduating should be considered dropouts.

By doing so, Greene incorrectly lowers the graduation rate by 5 percent overall, by 11 percent for Hispanics and by 18 percent for Asians. This is absurd. High schools have problems, but no serious observer would fault them for failing to graduate immigrants who may never have enrolled or enrolled only briefly.

A Mischaracterization

With my colleague Joydeep Roy, I've tried to develop more accurate graduation rates using data from the Census Bureau and from high-quality longitudinal studies that track the same students from earlier in their schooling until they graduate or drop out. The census includes those who got GEDs as high school graduates, but we correct for this by subtracting the number of GEDs awarded from census graduation rates.

The best longitudinal study, the National Education Longitudinal Study, surveyed 8th graders and then followed them into high school and young adulthood. If respondents told surveyors they had graduated from high school, the surveyors checked the answers against high school transcripts to ensure the information was correct.

We have concluded that three-fourths of black students actually graduate with regular diplomas. It is a similar story for Hispanics. More than 80 percent of all students earn regular diplomas.

Does it really matter whether the overall graduation rate is about 80 percent, as we estimate, or 67-70 percent, as Greene and Swanson claim? After all, even a 20 percent dropout rate is too high.

Touting a 50 percent graduation rate for minorities is not only factually incorrect, but it mischaracterizes black students as too hard to reach or disinterested in education. In contrast, showing that 75 percent of black youth get diplomas and another 13 percent get GEDs is both factually correct and appropriately acknowledges the striving and persistence of black students.

Artificially low graduation statistics also encourage an across-the-board indictment of schools, leading to misguided or wrongfully targeted reform efforts--a little like sending homeland security funds to Montana and Idaho rather than to New York and Washington.

Socioeconomics Dictate

In truth, the dropout problem is concentrated in about 20 percent of high schools. Dropouts are only 3 percent of students from the upper three-fifths of the socioeconomic scale. (This calculation includes GED recipients as high school completers.) Yet the bottom fifth has 27 percent who fail to complete high school in any way, and 38 percent who fail to achieve a regular diploma. The next higher fifth has 13 percent who don't complete high school.

Interestingly, among students in the lowest socioeconomic fifth, blacks have a higher probability of graduating than do low-income whites or Hispanics. This is a reminder of how intertwined race and class are in our society. To the extent we have a dropout crisis, it is significantly a crisis of youth at the bottom of the socioeconomic scale, regardless of race.

Lawrence Mishel, an economist, is president of the Economic Policy Institute (www.epi.org). E-mail: [e-mail]. He is the co-author of Estimating Graduation Rates and Trends.